Spleen colony formation (Till and McCulloch, 1961) by cells collected from mice treated with the cytotoxic drug 5-fluorouracil (5-FUBM) differs from spleen colony formation by normal bone marrow cells (NBM). The number of spleen colonies formed after the graft of 5-FUBM rises with the passage of time since the graft (extra spleen colonies). It does not after the graft of NBM. Hodgson and Bradley (1979) proposed that extra spleen colonies might form in this way: a progenitor of CFUs ('the pre-CFUs') seeded preferentially to the marrow of recipient mice. It formed CFUs there which migrated to the spleen and formed extra colonies. This hypothesis, if it were true, would support another hypothesis, the generation-age hypothesis (Rosendaal et al., 1976, 1979), which proposes that the number of divisions there have been in the history of a stem cell is one determinant of the order in which they start cycling to form other stem cells and differentiated cells. We investigated four possible sources of extra colonies. They might arise by reseeding of CFUs formed in the marrow or the spleen; they might arise from CFUs which settle in spleen but require a variable time until they can start to cycle, or they might arise after a variable time as CFUs repair the damage done them by 5-FU. We have established that the CFUs which eventually form spleen colonies are in the spleens of recipients 24 hours after the graft. Extra colonies too, do not arise by reseeding of CFUs formed in the spleens of recipients. It is possible that extra colonies arise as CFUs repair the damage done by 5-FU.